Specifically adsorbed ferrous ions modulate interfacial affinity for high-rate ammonia electrosynthesis from nitrate in neutral media

Significance

Nitrogen is an essential element for life. Ammonium, compared with nitrate, is preferred by crops to facilitate the rapid synthesis of amino acids and is of critical importance to the productivity of terrestrial ecosystems worldwide. Therefore, the synthetic ammonia industry has developed rapidly, but also consumes global 1 to 2% of energy and contributes 1 to 2% CO₂ emission. In this study, the ambient ubiquitous iron oxide was used to develop an electrochemical nitrate reduction to ammonia and obtain an extremely high NH₃ yield rate under neutral pH conditions with no CO₂ emissions and low energy consumption. These findings highlight a unique way for ammonia fertilizer synthesis from ambient nitrate, thereby showing significant impact on fertilizer industry, groundwater nitrate remediation, and climate change.

Abstract

The electrolysis of nitrate reduction to ammonia (NRA) is promising for obtaining value-added chemicals and mitigating environmental concerns. Recently, catalysts with high-performance ammonia synthesis from nitrate has been achieved under alkaline or acidic conditions. However, NRA in neutral solution still suffers from the low yield rate and selectivity of ammonia due to the low binding affinity and nucleophilicity of NO₃⁻. Here, we confirmed that the in-situ-generated Fe(II) ions existed as specifically adsorbed cations in the inner Helmholtz plane (IHP) with a low redox potential. Inspired by this, a strategy (Fe-IHP strategy) was proposed to enhance NRA activity by tuning the affinity of the electrode–electrolyte interface. The specifically adsorbed Fe(II) ions [SA-Fe(II)] greatly alleviated the electrostatic repulsion around the interfaceresulting in a 10-fold lower in the adsorption-free energy of NO₃⁻ when compared to the case without SA-Fe(II). Meanwhile, the modulated interface accelerated the kinetic mass transfer process by 25 folds compared to the control. Under neutral conditions, a Faraday efficiency of 99.6%, a selectivity of 99%, and an extremely high NH₃ yield rate of 485.8 mmol h⁻¹ g⁻¹ _FeOOH were achieved. Theoretical calculations and in-situ Raman spectroscopy confirmed the electron-rich state of the SA-Fe(II) donated to p orbitals of N atom and favored the hydrogenation of *NO to *NOH for promoting the formation of high-selectivity ammonia. In sum, these findings complement the textbook on the specific adsorption of cations and provide insights into the design of low-cost NRA catalysts with efficient ammonia synthesis.

Ammonia (NH₃) is pivotal to the fertilizer industry and is one of the most globally important chemicals (1–4). NH₃ is mainly produced through Haber–Bosch process, consuming 1 to 2% of global energy and releasing 1 to 2% of the global anthropogenic CO₂ emission (5, 6). Every year, approximately 500 million tons of NH₃ are produced to meet the needs of N-based fertilizer production (7). Nitrate is ubiquitous in water bodies altered by human activities, thereby threatening the human health and ecology balance (8–11). Ammonia or nitrogen gas (N₂) from nitrate electroreduction serves as a mild strategy to obtain value-added chemicals and environmentally friendly products (12, 13). The number of electrons transferred for ammonia and N₂ reduction requires 8 and 5 electrons, respectively. Obviously, the chemical properties of the two products are different. N₂ is a harmless gas that exists widely in the atmosphere, and ammonia is an important chemical product as well as the main raw material for fertilizers (9, 14). Therefore, the conversion of waste aqueous nitrate into value-added NH₃ is highly desirable from an environmental and energetic viewpoint.

The electrocatalytic reduction of nitrate reduction to ammonia (NRA) could solve the energy and environmental problems without the release of CO₂. Therefore, various noble metals, alloys, and transition metal oxides, such as Cu-based electrocatalysts and Fe-based electrocatalysts, have been used for NRA (15–17). However, NRA still suffers from low yield rate and selectivity of NH₃ in neutral solutions. This is mainly due to the low binding affinity and nucleophilicity of NO₃⁻ caused by the planar symmetrical (D3h) resonant structure besides the complicated eight-electron coupled nine-proton transfer process (18, 19). Moreover, the electrostatic repulsion at the electrode–electrolyte interface limits the mass transfer of NO₃⁻, sharply dropping the NO₃⁻ concentration. For typical interface reactions, NRA consists of two basic steps: adsorption and catalysis, in which adsorption starts the NRA, and hence considered a key factor for overcoming the constraints of low binding capacity of NO₃⁻ under neutral conditions. Therefore, a rational design of the electrochemical catalytic interface plays an important role in promoting binding and mass transfer of NO₃⁻. Currently, several strategies have been successfully used, including nano/micro-confinement and construction of internal electric fields exhibiting excellent performances by accumulating NO₃⁻ around the surface region (20, 21). However, unfortunately, great challenges remain in neutral solutions to promote the yield rate and selectivity of NH₃.

Redox reactions of Fe(II) and Fe(III) are commonly found in nature and play important roles in controlling biogeochemical cycle of natural sediments, soils, and aquifers (22). In particular, Fe(II) associated with iron mineral phases, such as goethite (α-FeOOH), akaganeite (β-FeOOH), and lepidocrocite (γ-FeOOH), forms highly reactive adsorbed Fe(II) species through chemical adsorption or electron transfer, leading to elevated redox reactivity by lowering the standard redox potential when compared to that of solution (23, 24). The adsorbed Fe(II) is a key factor for determining chemically and microbially driven electron transfer processes across iron mineral–water interfaces (25). Importantly, the external electrons could undergo reduction reactions with Fe(III) on FeOOH surface to release generated Fe(II) from the lattice existing as adsorbed Fe(II) at the FeOOH–water interface (25, 26). In nature, such a whole process occurs in a neutral aqueous solution (27). Previous studies have reported this adsorbed Fe(II) ability of reducing metal ions and inorganic salt ions due to the ubiquitous presence of iron in lithosphere, as well as the wide range of reduction potentials coupling Fe(II)/Fe(III) (28–30).

Herein, based on electrical double-layer model, we found that Fe(II) generated from FeOOH binds to the surface via specific adsorption at the inner Helmholtz plane (IHP) and exhibits a high redox reactivity. Thus, a strategy employing the specifically adsorbed Fe(II) [SA-Fe(II)] ions at the IHP (Fe-IHP strategy) was proposed for tuning electrode–electrolyte interface affinity. The Fe-IHP strategy effectively increased the NO₃⁻ number density by three folds at 0.3 nm away from the surface with modulated interface affinity confirmed by molecular dynamics (MD) simulation. Furthermore, the SA-Fe(II) ions subtly weakened the electrostatic repulsion near the surface, thereby accelerating the NO₃⁻ mass transfer and leading to a 10-fold higher adsorption capacity than in the absence of it. In neutral aqueous media, the Faraday efficiency (FE) of ammonia reached 99.6% and selectivity was 99% (45 mg L⁻¹, NO₃⁻–N) with a high yield rate of 485.8 mmol h⁻¹ g⁻¹ _FeOOH (0.2 mol L⁻¹ NO₃⁻–N). Moreover, the kinetic constant was 25-fold higher than that without Fe(II), suggesting accelerated kinetic mass transfer processes with the Fe-IHP strategy. In sum, insights into the design of universal low-cost NRA catalysts with high efficiency were provided, promising for fertilization and global groundwater nitrate remediation.

Results and Discussions

As a typical interfacial reaction, the first step of NRA is adsorption which efficiently improves the NO₃⁻ accumulation (28, 31, 32). However, the strong electrostatic repulsion around the surface in a neutral solution would not be conducive to NO₃⁻ adsorption (Fig. 1 A, Left). Inspired by the stable existence of Fe(II) ions adsorbed on the surface of FeOOH in natural aqueous bodies, we proposed a strategy employing the in-situ-generated Fe(II) ions in the IHP (Fe-IHP strategy) to enhance interface interactions for NRA (Fig. 1 A, Right). This induced a local microenvironment with weak electrostatic repulsion or even created electrostatic attraction, leading to a suitable interface microenvironment for NO₃⁻ adsorption. Computational fluid dynamics (CFD) simulations were performed to investigate the diffusion rate of adsorbed Fe(II) on NRA reaction, which further reveal the role of adsorbed Fe(II). To this end, a simple array structure model was constructed for CFD simulations (SI Appendix, Fig. S1). As shown in Fig. 1B, the concentration of adsorbed Fe(II) at the surface tending to zero indicated a complete utilization and led to a decline in NO₃⁻ concentration, while NO₂⁻ and NH₄⁺ gradually rose during the simulation time of 5 s. Furthermore, the enhancement in the diffusion rate of adsorbed Fe(II) led to a significant decrease in the reduction rate of NO₃⁻, as well as a slower generation rate of NH₄⁺ (SI Appendix, Fig. S2). The real-time graphs of NH₄⁺ concentration suggested no obvious accumulation near the surface region of the array structure, indicating faster diffusion of NH₄⁺ (SI Appendix, Fig. S3). Thus, according to CFD results, the retention time of adsorbed Fe(II) at the surface promoted a high reactivity of NRA. Obviously, this form of in-situ generation of adsorbed Fe(II) would prolong its retention time at the surface.

Fig. 1.

(A) Scheme of Fe-IHP strategy. (B) CFD simulations of NO₃⁻, NH₄⁺, and NO₂⁻ concentrations as a function of simulation time under completely consumed adsorbed Fe(II). (C) XRD patterns of γ-FeOOH, Fe₃O₄, FeO, and SS. (D) SEM images. (E) TEM image (Inset represents SAED, Scale bar: 5 1/nm.) (F) HRTEM image of NA-FeOOH. (G) Mapping of Fe and O elements (Scale bar: 200 nm.) (H) XPS of Fe species and (I) Raman spectra. (J) ⁵⁷Fe Mössbauer spectrum of NA-FeOOH.

FeOOH was selected as the source for the in-situ generation of Fe(II) ions. Based on the large specific area of the array structure, a three-dimensional nanoarray structure FeOOH (NA-FeOOH) was synthesized on the stainless steel (SS) substrate via a simple electrochemical deposition method. By changing the deposition time, two unevenly distributed FeOOH electrodes (AS1-FeOOH and AS2-FeOOH) were synthesized as controls (SI Appendix, Fig.  S4). As shown in Fig. 1C, the X-ray diffraction (XRD) analysis showed peaks for NA-FeOOH at 14.1, 27.1, 36.5, 38.1, 46.9, 52.7, 60.7, and 64.9°, belonging to γ-FeOOH (JCPDS: 44-1415). The characteristic peaks at ~30.1 and 35.6° confirmed the existence of the Fe₃O₄ phase (JCPDS: 79-0417). Apart from Fe₃O₄, the peak at 60.3° was ascribed to FeO (JCPDS:74-1886), indicating another source of Fe(II) species. In addition, the three background peaks at 43, 52, and 75° were attributed to the SS substrate with main components shown in SI Appendix, Table S1. To further confirm the strong peak at 43° of SS, the XRD patterns were obtained on three parallel substrates (SI Appendix, Fig. S5), and the results were consistent with those of the SS in Fig. 1C. Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) confirmed the sheet-like array structure of NA-FeOOH uniformly distributed on SS with a thickness of 8 to 12 nm (Fig. 1D). Selected area electron diffraction (SAED) and high-resolution TEM (HRTEM) displayed NA-FeOOH made of an overall polycrystalline structure with a partially exposed single crystal (301) plane (Fig. 1 E and F). The elemental mapping revealed that Fe and O were uniformly distributed on the nanosheets (Fig. 1G).

The X-ray photoelectron spectroscopy (XPS) spectra of the Fe element confirmed the coexistence of Fe(0), Fe(II), and Fe(III) species. As shown in Fig. 1H, the Fe 2p XPS spectrum of NA-FeOOH showed three sets of peaks at 709.1, 722.6 eV; 710.3, 724.0 eV; or 711.8, 725.6 eV, corresponding to Fe(0), Fe(II), or Fe(III) species in Fe 2p3/2 or Fe 2p 1/2 regions, respectively (16, 33, 34). For accurate results, the XPS spectra of the other two parallel samples were measured. As depicted in SI Appendix, Fig. S6, the positions of the three Fe species were consistent with those in Fig. 1H. Accordingly, the calculated atomic proportions of Fe(0), Fe(II), and Fe(III) species ranged from 23.2 to 24.2%, 24.7 to 26.2%, and 50.6 to 51.3%, respectively (SI Appendix, Table S2). The appearance of Fe(0) might be attributed to the overreduction in iron oxide or iron oxyhydroxide during the electrodeposition process. Meanwhile, Raman spectroscopy confirmed the characteristic peaks of γ-FeOOH at 247 and 373 cm⁻¹, as well as Fe₃O₄ at 650 cm⁻¹ (Fig. 1I) (34–36). To further explore the valence state and phase of NA-FeOOH, Mössbauer spectroscopy was performed and the data are displayed in Fig. 1J. Combined with the XRD spectra, the orange area was confirmed as γ-FeOOH phase with a mass ratio of ~60% (29). Meanwhile, the green and yellow areas were considered as Fe₃O₄ phase with a mass ratio of ~28% (26). In addition, Fe(0) species (pink and purple areas) also existed (37).

Benefiting from the weakened interfacial electrostatic repulsion and the direct exposure of adsorbed Fe(II) ions, features such as the selectivity (S_NH4+), FE, and yield rate of NH₃ (Y_NH4+) reached values of 99%, 99.6%, and 71.1 mmol h⁻¹ g⁻¹_FeOOH, respectively, in neutral solution (−0.67 V vs. RHE, 45 mg L⁻¹ NO₃⁻-N) (Fig. 2A). Furthermore, the Y_NH4+ attained an extremely high NH₃ yield rate of 485.8 mmol h⁻¹ g⁻¹_FeOOH (4.96 mg h⁻¹ cm⁻²) at 0.2 mol L⁻¹ NO₃⁻-N within 1 h as shown in SI Appendix, Table S3, when compared to that of other works (SI Appendix, Table S4). The image of color reaction, original data of NH₄⁺-N determination, and I-t curves at the time of the reaction are enclosed in SI Appendix, Figs. S7 and S8. In Fig. 2B, compared to the SS electrode (black line), the LSV curves of the NA-FeOOH with the array structure illustrated a significant reduction peak (blue and red lines), indicating the occurrence of NRA reaction. The performances of the Fe-IHP strategy at extremely low concentration were verified by further selecting the concentration of 10 mg L⁻¹ NO₃⁻-N as proposed by the U.S. Environmental Protection Agency for control of drinking water.

Fig. 2.

(A) FE, selectivity of NH₄⁺ (S_NH4+), and yield rate of NH₄⁺ (Y_NH4+) under various voltages (vs. Ag/AgCl). (B) The LSV curves of NA-FeOOH in 45 (blue line) and 135 mg L⁻¹ NO₃⁻–N (red line) and SS (black line) in 45 mg L⁻¹ NO₃⁻–N. (C) S_NH4+ and NO₃⁻ reduction efficiency (R_NO3−) values by NA-FeOOH at 10 and 45 mg L⁻¹ NO₃⁻–N. (D) Concentrations of NO₃⁻–N, NO₂⁻–N, and NH₄⁺–N during the reaction. (E) R_NO3− and Y_NH4+ after adding TCD, Na-EDTA, 1,10-phenanthroline hydrate (1,10 PH) for Fe(II) complexion and (F) pseudo first-order kinetics reaction fitting curves by NA-FeOOH. Note that all the experiments are carried out in 0.1 M PBS buffer with 0.5 M Na₂SO₄.

As shown in Fig. 2C, the NA-FeOOH electrode exhibited 95% and 97% selectivity and NO₃⁻–N reduction efficiency within 1 h, indicating that the accumulation of NO₃⁻ ions occurred even at extremely low NO₃⁻–N concentration levels. Under neutral conditions (Fig. 2D), the concentrations of NO₃⁻–N (45 mg L⁻¹) almost reduced to 0 (99% reduction efficiency) within 60 min, and that of NH₄⁺–N was close to 45 mg L⁻¹ (~100% conversion efficiency). In particular, the maximum ratio of toxic NO₂⁻–N was lower than 2.6%. To evaluate the role of NA-FeOOH, we performed control experiments using SS substrates and control FeOOH electrodes. The reduction efficiency of NO₃⁻–N reached only 40% and 46% by the AS1-FeOOH and AS2-FeOOH electrodes, respectively (SI Appendix, Fig. S9). In addition, the concentration of NO₃⁻–N decreased from 45 to 37.1 mg L⁻¹ with the reduction efficiency of 17.4%, suggesting that NA-FeOOH was the source of active species in nitrate reduction (SI Appendix, Fig. S10). The concentrations of NO₃⁻–N, NO₂⁻–N, and NH₄⁺–N were determined by UV-vis spectrophotometry with the standard concentration–absorbance curves (SI Appendix, Figs. S11–S13). Note that all experiments were performed under Ar-statured conditions in an H-type reactor with attached digital images and detailed parameters (SI Appendix, Fig. S14).

To further explore the role of adsorbed Fe(II) on NRA, three Fe(II) scavengers were selected for quenching experiments. The concentration of these three scavengers was determined as 4 mM according to the results of quenching with Na-EDTA (SI Appendix, Fig. S15). As shown in Fig. 2E, the addition of trisodium citrate dihydrate (TCD), Na-EDTA, and 1,10-phenanthroline hydrate (1,10 PH) resulted in an immediate decrease in NO₃^––N reduction efficiency to 19%, 21%, and 14.5%, respectively. These values were almost five-, five-, and sixfold lower than those without added scavengers, respectively. Furthermore, the Y_NH4+ reached 5.4, 6.1, and 7.6 mmol h⁻¹ g⁻¹_FeOOH, respectively, corresponding to a decline by one order of magnitude than without scavengers (71.1 mmol h⁻¹ g⁻¹_FeOOH). Since atomic H (H*) could also be used as an active species for NO₃⁻–N reduction, tertiary-butanol (TBA) was employed as a scavenger for radical quenching experiments. The results suggested no significant change in NO₃⁻–N reduction efficiency and kinetic constant with the increase in the concentration of TBA (SI Appendix, Fig. S16). Moreover, the in-situ ESR experiments did not detect the characteristic signals of H* (SI Appendix, Fig. S17). Electrocatalytic NO₃⁻–N reduction corresponded well with the pseudo first-order kinetics (16, 34). As shown in Fig. 2F, the |k| value reached 0.058 min⁻¹, a value almost 25-fold higher than those obtained in the presence of scavengers (0.0021 to 0.0023 min⁻¹) (SI Appendix, Fig. S18 and Table S5), indicating that the adsorbed Fe(II) ions accelerated the reaction kinetics. Therefore, the adsorbed Fe(II) was the dominant active species involved in NRA. Besides, the excellent NO₃⁻–N reduction efficiency and selectivity of NH₄⁺–N after five cyclic stability tests of NA-FeOOH confirmed the feasibility of the Fe-IHP strategy (SI Appendix, Fig. S19). Furthermore, the morphology and Fe element valences of NA-FeOOH after NRA reaction were investigated by SEM and XPS. SEM image demonstrated a uniformly distributed nanoarray structure after NRA reaction. The XPS analysis demonstrated that Fe(0), Fe(II), and Fe(III) species of NA-FeOOH coexisted with the relative atomic ratio of 30.5%, 25.3%, and 44.2%, respectively, indicating a relative stability of the catalyst (SI Appendix, Fig. S20).

The adsorption of Fe(II) ions on the electrode surface would theoretically take place in two forms. The first was in the form of nonspecifically adsorbed Fe(II) [non-SA-Fe(II)] ions in the form of solvated Fe(II) ions, and the second was in the form of specifically adsorbed Fe(II) [SA-Fe(II)] ions if present. To clarify the key Fe species driving NRA, the location, concentration, and role of adsorbed Fe(II) ions were further investigated. Bard’s work clearly states that the electrical centers of the specifically adsorbed ions on solid surface and the nearest solvated ions in solution are called IHP and outer Helmholtz plane (OHP), respectively (38). To determine the location of the adsorbed Fe(II) ions, it is vital to confirm whether the specifically adsorbed Fe(II) ions occur on the surface. The potential of zero charge (PZC), defined as the potential where no excess charge exists at the electrode surface, is the direct way to determine whether specific adsorption occurs to a surface (38). PZC could be determined by the minimum differential capacitance curve in a dilute electrolyte (39, 40). As shown in Fig. 3A, with the open-circuit potential (OCP, −0.5 V, SI Appendix, Fig. S21) as the initial potential, the potential at the lowest point which represented the PZC reached −0.23 V (red line) after Fe(II) ion adsorption. However, in the absence of adsorbed Fe(II) ions, the PZC was determined as −0.36 V (black line). An obvious positive shift of 0.13 V in PZC occurred after Fe(II) ion adsorption. According to the textbook by Bard (Chapter 13, Section 4.1), the positive shift of PZC indicates that cations are specifically adsorbed (38, 41, 42).

Fig. 3.

(A) Variations of PZC with no Fe(II) or under different concentrations of Fe(II) ions by the minimum differential capacitance (C_d) curves. (B) The surface Zeta potential before Fe(II) ions adsorption or immersed in 100 mM FeCl₂ solution measured in standard solution containing polystyrene latex suspended in an aqueous buffer. (C) Concentration and adsorption quantity of adsorbed Fe(II) ions with various ionic strengths (NaCl as the model solution). (D) Concentration of NO₃⁻–N in FeCl₂ solution (Fe(II): 200 mg L⁻¹) with time under neutral condition. (E) Proposed transfer pathways and existing forms of the generated Fe(II) ions at FeOOH surface based on the specific adsorption. (F–H) Overlay images of adsorbed Fe(II) ions after immersion in 10 mM, 100 mM, and 800 mM NaCl solution, or (I) without Fe(II) ion adsorption measured by fluorescence microscopy (Scale bar: 100 µm).

To identify the effects of solvated Fe(II) ions, PZC was further determined in a series of Fe(II) concentrations by the minimum differential capacitance. As shown in Fig. 3A, the increase in Fe(II) concentration from 1 to 100 mM led to a positive shift in PZC from −0.20, −0.15, −0.12, −0.06 and +0.05 to +0.10 mV, with change from negative to positive values. This could be ascribed to the higher Fe(II) ion electrolyte, which induced more SA-Fe(II) ions on the electrode surface, leading to countercharge on the electrode due to more polarization than that in the solution. To regain the condition σ^м = 0 (σ^м representing the excess charge in metal phase at the interface), the potential must be shifted to a more positive value (39). The excess charge of SA-Fe(II) ions could then be exactly counterbalanced by the opposite charge in the diffuse layer. Another indicator of specific adsorption of charged species is the Esin–Markov effect, which is manifested by a shift in the PZC as a function of the change in electrolyte concentration (39). Our results suggested that PZC shifted positively with the increase in Fe(II) ions (Fig. 3A), consistent with the Esin–Markov effect and further confirming the presence of the specifically adsorbed Fe(II) ions at the electrode surface. Some studies reported that surface Zeta potential would move positively once a specific adsorption of cation occurs (43, 44). As shown in Fig. 3B, the surface Zeta potential before adsorption was determined to be −10.5 ± 2.04 and further increased to 12.8 ± 1.95 mV after immersion in 100 mM FeCl₂ solution, suggesting the occurrence of Fe(II) ion adsorption on the electrode surface.

To confirm the coexistence and concentrations of SA-Fe(II) and non-SA-Fe (II) ions, ionic strength experiments were performed using NaCl solution. It was reported that the non-SA-Fe(II) ions would be gradually replaced by other cations (such as Na⁺) with the increase in ionic strength (45, 46). However, the SA-Fe(II) ions were hardly disturbed by ionic strength according to the fundamental knowledge that metal ions could only stay at OHP rather than that at IHP (39). In addition, considering that specific adsorption is often accompanied by the formation of chemical bonds, it is difficult for Na⁺ ions to replace metal ions that have formed chemical bonds. Therefore, the SA-Fe(II) ions were basically equal to the adsorbed Fe(II) ions with little change in concentration at higher ionic strength, and then the non-SA-Fe(II) ions could be further calculated under the condition of a specific ionic strength. As shown in Fig. 3C (column) and SI Appendix, Table S6, the total adsorbed Fe(II) ions decreased from 133.1 ± 25 to 37.2 ± 7 µg L⁻¹ as the NaCl concentration increased from 0 to 1,000 mM, indicating the existence of non-SA-Fe(II) ions. Moreover, the concentration of adsorbed Fe(II) ions stabilized at 37.2 ± 7 µg L⁻¹ at a NaCl concentration of 800 mM. Therefore, the concentration of the SA-Fe(II) ions was estimated as 37.2 ± 7 µg L⁻¹. Meanwhile, the adsorption quantity of Fe(II) ions varied from 221.9 ± 43 to 61.9 ± 12 µg g⁻¹_FeOOH (Fig. 3C, green line). The concentration of non-SA-Fe(II) ions obtained under different ionic strength conditions is summarized in SI Appendix, Table S7. Note that the ionic strength experiments only provided estimates since the non-SA-Fe(II) ions could not be completely replaced by other cations.

To explore the role of the solvated Fe(II) ions in NRA, the concentration of NO₃⁻–N in Fe(II) solution with time was determined under Ar-saturated condition in neutral media. As shown in Fig. 3D, there was almost no change in the concentration of NO₃⁻–N within 3 h, suggesting that the solvated Fe(II) ions could not reduce nitrate to ammonium. As shown in Fig. 3E, based on the special adsorption, these in-situ-generated Fe(II) ions mainly existed in the form of SA-Fe(II) in the IHP and solvated Fe(II) ions in the OHP and diffuse layer (47). Furthermore, the SA-Fe(II) ions at the IHP were responsible for the high-rate electroreduction of nitrate to ammonium as the key active species. As shown in Fig. 3 F–I, the brightness variation of the overlay images by the fluorescence microscopy further confirmed the existence of non-SA-Fe(II) ions at the FeOOH surface. Moreover, the increase in the concentration of NaCl from 10 to 800 mM led to a decline in the fluorescence intensity of Fe (II) ions, indicating Na⁺ in the solution subjected to ion exchange reaction with the non-SA-Fe (II) ions on the FeOOH surface, but it could not replace the SA-Fe(II) ions. The fluorescence and bright field images in the absence of Fe(II) ion adsorption or adsorbed Fe(II) ions after immersion in 10 mM, 100 mM, and 800 mM NaCl solution are supplemented in SI Appendix, Fig. S22. X-ray absorption fine structure (XAFS) and Fourier-transform extended X-ray absorption fine structure (FT-EXAFS) spectra were carried out to probe the coordination states of the Fe element (5, 10). The XAFS and FT-EXAFS data confirmed the increased intensity of Fe(II)–O and Fe(III)–O bonds after Fe(II) ion adsorption (SI Appendix, Fig. S23), indicating the presence of SA-Fe(II) ions at the FeOOH surface. Our findings complement the textbook on the specifically adsorbed metal cations.

In-situ Raman spectroscopy was carried out to further reveal the mechanism of NRA. As shown in Fig. 4A, the intensity of the characteristic peak in γ-FeOOH at 247 cm^–1 gradually decreased as the reaction progressed, while the characteristic peak of β-FeOOH at 320 cm^–1 gradually increased. This suggested the occurrence of interface reconstruction with the main phase partially transformed from γ-FeOOH to β-FeOOH. In addition, the characteristic peaks at 247 and 320 cm^–1 revealed an increase in the peak ratio of β-FeOOH to 41% at 5 min followed by stability at ~80% from 30 min until the end of the process (SI Appendix, Table S8), indicating β-FeOOH as the main phase during NRA. TEM and HRTEM images of NA-FeOOH electrode after 30 min reaction confirmed locally exposed (001) plane of β-FeOOH after the reconstruction (SI Appendix, Fig. S24), even though the mixed crystal phase still existed. Moreover, density functional theory (DFT) determined bulk formation energies of γ-FeOOH and β-FeOOH as ΔE_f = −0.17 eV and ΔE_f = 0.29 eV, respectively. This suggested that the surface reconstruction weakened the high energy barrier, thereby forming a more stable β-FeOOH phase with the low-index exposed plane of (001) (SI Appendix, Fig. S25).

Fig. 4.

(A) In-situ Raman spectra of NA-FeOOH during the whole reaction. (B) Concentration and redox potential of adsorbed Fe(II) ions under different applied voltages as well as no voltage (reaction time, 30 min, 2 cm²). (C) Concentration of adsorbed Fe(II) as a function of reaction time (2 cm²; Inset shows the scheme of (301) plane of γ-FeOOH transformed to the (001) plane of β-FeOOH). (D) MD simulations of the (001) plane of β-FeOOH in NaNO₃ solution. (E) Number of real-time NO₃⁻ ions along the z-axis from the electrode surface based on MD simulations. (F) Proposed three steps of NRA based on the Fe-IHP strategy.

The thermodynamic behaviors of the interfacial electron were investigated by determining the redox potential of adsorbed Fe(II) ions at the electrode surface using anthraquinone-2,6-disulfonate (AQDS) as a redox probe according to the Nernst equation. Each voltage illustrated a stable ratio of AQDS in the reduced state [AQDS (red)]/AQDS and oxidized state [AQDS (ox)] at the thermodynamically stable state with different color degrees (SI Appendix, Fig. S26). The ratio then gradually increased during the next thermodynamically stable state as the voltage rose (SI Appendix, Fig. S27 and Table S9). As shown in Fig. 4B (blue lines), the redox potential gradually stabilized at −183 mV at the applied voltage of −1.3 V, a value 29 mV lower than the initial potential (−154 mV). Thus, the reduction ability of the adsorbed Fe(II) ions improved after reconstruction. The cyclic voltammetry curves showed an increase in the oxidation peak area, while the reduction potential of NO₂⁻-N shifted positively after the reconstruction, indicating an enhanced NRA reaction (SI Appendix, Fig. S28). Furthermore, CFD simulations demonstrated the importance of the diffusion rate of adsorbed Fe(II) on NRA, while the concentration was the main determining factor. In addition, the concentration of adsorbed Fe(II) was measured by ICP-MS.

As depicted in Fig. 4B (columns), the concentration reached the highest value of 61.4 ± 5.5 µg L⁻¹ cm⁻² under −1.3 V condition, a value thrice that of the initial state without applied voltage (SI Appendix, Table S10). Moreover, the concentration of adsorbed Fe(II) on the NA-FeOOH surface showed an increase followed by a decrease, with the highest value reached after 30 min of reaction (Fig. 4C and SI Appendix, Table S11). The tunnel configuration of the (001) plane in β-FeOOH was estimated to 5.0 Å, a value almost twice larger than that of the (301) plane of γ-FeOOH (2.7 Å), which was conducive to more NO₃⁻ ion adsorption (Fig. 4 C, Inset image).

To accurately quantify the dynamic behaviors of NO₃⁻ ions near the surface region, MD simulations were performed based on the (001) plane of β-FeOOH (SI Appendix, Fig. S29). In the absence of Fe(II), NO₃⁻ ions near the surface looked dispersed at a low number density (SI Appendix, Fig. S30). However, a significant accumulation effect of NO₃⁻ ions near the surface was noticed in the presence of Fe(II) ions. This can more prominently be observed in the enlarged image (Fig. 4D). Furthermore, the maximum number of NO₃⁻ within a distance (10 Å) reached 10 ions /Å^-3 at ~3 Å away from the surface, a value threefold higher than that without Fe(II) (Fig. 4E). Meanwhile, the real-time concentration of NO₃⁻ increased rapidly after the start of the reaction and remained stable at ~ 23 ions/Å⁻³, while the control without Fe(II) was only 13 ions/Å⁻³ (SI Appendix, Fig. S31). To be mentioned, the number density of Na⁺ decreased significantly in the range of 10 to 20 Å, which was due to the neutralization of the negative charges at the NA-FeOOH surface by the SA-Fe(II) ions (SI Appendix, Fig. S32).

Overall, the MD simulation results proved the accumulation effect of NO₃⁻ ions caused by the adsorbed Fe(II) ions. Therefore, a NRA pathway was proposed based on the Fe-IHP strategy. This can be divided into three steps: i) NO₃⁻ adsorption, ii) NO₃⁻ accumulation, and iii) NO₃⁻reduction (Fig. 4F). This looked very different from the traditional two steps consisting of NO₃⁻ adsorption and reduction. The key step of this “three-step” strategy consisted of generating a layer of SA-Fe(II) ions at the surface, leading to the accumulation of negatively charged NO₃⁻ ions. Compared to the traditional adsorption–reduction model, the proposed Fe-IHP strategy showed three advantages. First, this can effectively strengthen the interface affinity under neutral conditions. Second, the SA-Fe(II) ions reacted directly with NO₃⁻ as the active species. Third, the accelerated electron transfer between SA-Fe(II) and SA-Fe(III) ions improved the kinetic reaction process of NRA.

To rationalize the excellent performances, density functional theory (DFT) calculations were conducted on the (001) plane of β-FeOOH (SI Appendix, Figs. S33 and S34). As shown in Fig. 5A, the adsorption energy of NO₃⁻ on the β-FeOOH surface was reduced to −0.984 eV, a value nearly 10-fold that of −0.102 eV without Fe(II). The predicted density of states (PDOS) for NO₃⁻ adsorption on the (001) plane of β-FeOOH revealed a large overlap with Fe(II) when compared to that without Fe(II), suggesting an enhanced adsorption capacity (SI Appendix, Fig. S35). The electrochemical reaction (NO₃⁻ + 6H₂O + 8e⁻ → NH₃+ 9OH⁻) was represented by a series of deoxidation reactions consisting of (*NO₃ → *NO₂ → *NO →*N), followed by a series of catalytic hydrogenation reactions (*N → *NH → *NH₂ → *NH₃) (20, 48, 49). As shown in Fig. 5B, the sequence of deoxidation and hydrogenation revealed NRA involving three different pathways by forming the intermediate *N or *NOH (SI Appendix, Fig. S36). In N₂-pathway, the Gibbs free energy (ΔG) of deoxidation from *NO to *N was estimated to be +0.41 eV. However, the ΔG of *NO which hydrogenated to *NOH required only −0.70 eV, suggesting a thermodynamically energy-favorable reaction. Furthermore, the high free energy change from *N to *N+*N (+0.4 eV), as well as subsequent *N+*N to *N₂ (+0.74 eV), significantly suppressed the N₂-pathway. After the formation of *NOH, hydrogenation and deoxygenation were two options for subsequent reactions. The free energy required for deoxygenation was estimated to be only 0.12 eV and that for hydrogenation was 0.40 eV. At this time, deoxygenation was more favorable to the NRA reaction (named *NOH-O pathway) than hydrogenation (named *NOH-H pathway). Consequently, the transformation of *NO to *N/*NOH was the rate-limiting step determining the ammonium selectivity, while the transformation of *NOH to *NH/*NHOH was considered as the rate-limiting step for the whole NRA reaction. DFT calculations confirmed that SA-Fe(II) ions significantly modulated the formation barriers of the key intermediates, resulting in high selectivity.

Fig. 5.

(A) The adsorption energy and (B) reaction pathways for NRA on the (001) plane of β-FeOOH with Fe(II) ions. The key intermediates are marked in yellow shadow. (C) Differential charge density distribution of *NO₃⁻ and (D) *NO on the (001) plane of β-FeOOH with or without Fe(II) (* represents the active site). Electron deletion and accumulation are represented in yellow and green, respectively. (E) DOS of Fe d orbital and d band center before NO₃⁻ adsorption with or without Fe(II) ions.

As shown in Fig. 5C, the bonding configuration of NO₃⁻ changed from the original single Fe(III)-O to the double Fe(II)-O as shown by the differential charge density distribution. Meanwhile, the calculations of Milliken charge demonstrated that 0.53 e⁻ from Fe d orbital transferred to N 2p orbital, a value almost 1.2-fold than that without Fe(II). Hence, significant improvements in the bonding capacity took place. Such double bonding configuration was mainly derived from the shortened bond length of Fe(II)-Fe(II) ions (0.329 nm) when compared to that of Fe(III)-Fe(III) atom (0.543 nm). To probe the origin of the energy barrier that led to high NH₃ selectivity, the differential charge density distribution of NO on the (001) plane of β-FeOOH was investigated (Fig. 5D). The electron-richer state of Fe(II) greatly suppressed the electron donation from NO while facilitating the electrons transferred to the p orbital of N atoms, further promoting the hydrogenation of *NO. This could be due to the accumulated electrons transferred to the N atom (0.47 e⁻), which was approximately threefold than that without Fe(II) (0.16 e⁻). This could explain why it had a much lower formation barrier for *NOH. As shown in Fig. 5E, the calculated d-band center values were −2.53 and −2.30 eV for β-FeOOH with and without adsorbed Fe(II), respectively. Since the double Fe(II)-O bonding configuration ensured a better adsorption capacity, the lower d-band center indicated more stable Fe(II) (48, 50, 51). DFT calculations confirmed the promotion of the interfacial NO₃⁻ accumulation, as well as a high yield rate of ammonia by the double bonding configuration and electron donated to N atoms.

Conclusions

We confirmed the existence of the specially adsorbed Fe(II) ions at FeOOH surface and demonstrated the key role of it in NRA. The enhanced interface affinity and the double bonding configuration induced by SA-Fe(II) ions successfully activated NO₃⁻ by 10 folds for adsorption energy and accumulated three folds for the number density of NO₃⁻ ions. The proposed Fe-IHP strategy also accelerated the electron cycling with a kinetic constant 25 folds higher than that in the absence of SA-Fe(II) ions. Under neutral conditions, the FE of NRA reached 99.6% and selectivity was 99%, with a high NH₃ yield rate. Even at an ultra-low concentration of 10 mg L⁻¹ NO₃⁻–N, a 95% selectivity and 97% reduction efficiency were obtained. DFT and in-situ Raman spectra confirmed the donation of electrons of Fe(II) facilitated to p orbitals of N atoms, thereby favoring the transformation of *NO to *NOH and promoting the formation of ammonia. Our findings complement textbooks on specific adsorption of cations and provide unique insights into the design of universal NRA catalysts, promising for ammonium fertilization and global groundwater nitrate remediation.

Materials and Methods

Preparation of Catalyst.

Stainless steel (size 304, 100 mesh, thickness 0.07 mm, and diameter 0.19 mm) was used as the substrate. FeCl₂ 4H₂O, NaNO₃, NaCl, sodium sulfate (Na₂SO₄, 99%), acetone, and ethanol were all supplied by Alfa Aesar (ACS reagent). All reagents were used as received without further purification. Ultrapure water (18.2 MΩ·cm⁻¹) was obtained from a Milli-Q Plus system (Millipore) and utilized for all experimental preparations, washing/rinsing, and dilution. The nanoarray structure FeOOH (NA-FeOOH) was fabricated through an electrodeposition method in a standard three-electrode system using SS (20×30 mm), Pt plate (10×10 mm), and Ag/AgCl as working electrode, counter electrode, and reference electrode, respectively. An electrochemical cell (50 mL) contained 30 mL electrolyte of 0.1 M FeCl₂ 4H₂O electrolyte (no pH adjustment). The electrodeposition current was set at a constant current density of −10 mA cm^-2 and room temperature for deposition time of 60 s, 300 s, and 600 s. The whole electrodeposition process was continuously stirred at 300 rpm to make the solution uniform. After deposition, the obtained electrodes were rinsed with deionized water and dried at 80 °C to yield NA-FeOOH/SS. The loading of NA-FeOOH (300 s) was calculated as 0.6 mg cm⁻² by comparing the mass difference before and after the reaction.

Experiment Procedure.

The electrochemical NO₃⁻–N reduction experiments were conducted using NA-FeOOH (active area 5 cm²) as working electrode in the H-type double cell under Ar-saturated conditions. The concentration of DO was determined as ~0.01mg L⁻¹ by a WTW oxi/340i oxygen probe (WTW Company) (52–54). Each chamber was filled with 50 mL electrolyte for the reduction reaction. The electrolyte was composed of 0.1 M phosphate-buffered saline (PBS) (pH = 7) and 0.5 M Na₂SO₄ solution. Unless mentioned otherwise, NO₃^-–N concentration at 200 ppm (~45 mg L⁻¹ NO₃⁻-N) was added in the cathode chamber at room temperature (25 ± 1 °C). The NO₃⁻–N, NO₂⁻–N, and NH₄⁺–N were measured by UV spectrophotometry (UV-1800, HITACHI) by taking out 0.5 mL electrolyte every 30 min and diluting it to 5 mL to meet the detection range. Notably, when the initial concentration of NO₃⁻–N was 0.2 M (2.8 g L⁻¹), 0.05 mL of the sample was taken at each time and diluted to 5 mL with ultrapure water to form a 100-fold diluted solution. The details and standard concentration–absorbance lines are presented in experimental section (SI Appendix, Figs. S11–S13).

The percentage of nitrate reduction efficiency (R_NO3−%) was calculated according to Eq. 1:

$${\mathrm{R}}_{\mathrm{NO}3-}\%=\left({\mathrm{C}}_0{-\mathrm{C}}_{\mathrm{t}}\right){/\mathrm{C}}_0\times 100\%.$$ [1]

The NH₄⁺ selectivity (S_NH4+%) was obtained by Eq. 2:

$${\mathrm{S}}_{\mathrm{NH}4+}{=\mathrm{C}}_{\mathrm{t}}\left({\mathrm{NH}}_4^{+}\right)/\left({\mathrm{C}}_0{-\mathrm{C}}_{\mathrm{t}}\right)\times 100\%.$$ [2]

The yield rate of NH₄⁺ (µmol h⁻¹ g_cat⁻¹) was calculated following Eq. 3:

$${\text{Y}}_{NH4+}={\text{C}}_{\text{t}}\left(N{H}_4^{+}\right)\times \text{V}/{\text{M}}_{NH4+}\times \text{t}\times {\text{m}}_{cat}.$$ [3]

The Faradic efficiency (FE%) of the cathodic nitrate reduction was calculated using Eq. 4:

$$\mathrm{FE}\%=8\mathrm{F}\times {\mathrm{C}}_{\mathrm{t}}\left({\mathrm{NH}}_4^{+}\right)\times \mathrm{V}/\left({\mathrm{M}}_{\mathrm{NH}4+}\times \mathrm{Q}\right),$$ [4]

where C₀ (mg L⁻¹) and C_t (mg L⁻¹) are the initial NO₃⁻–N concentration and its concentration at time t, respectively. C_t (NH₄⁺) represents the NH₄⁺–N concentration at time t, m_cat refers to the mass of NA-FeOOH (0.6 mg cm⁻²), V is the electrolyte volume in the cathode compartment (50 mL), t denotes the reaction time (1 h), S is the active area (5 cm²), M_NH4+ represents the molar mass of NH₃, F is the Faradaic constant (96,485 C mol⁻¹), and Q refers to the total charge passing the electrode.

Determination of the PZC.

The PZC of the NA-FeOOH electrode was tested through minimum differential capacitance measurements in a three-electrode system consisting of a platinum foil as the cathode, an Ag/AgCl electrode as the reference electrode, and 5 mM NaCl solution as the electrolyte. The electrochemical impedance-potential spectra were measured at a frequency of 1 Hz with a sinusoidal voltage perturbation of 5 mV and a potential increment of 5 mV by an electrochemical workstation (CHI 660e). The initial potential was set at the OCP (−0.5 V, SI Appendix, Fig. S21), with electrode area of 2cm² (NA-FeOOH loading: 0.6 mg cm⁻²) under Ar-saturated condition. In order to explore the effects of the solvated Fe(II) ions, the PZC of the NA-FeOOH was determined by the minimum differential capacitance in a series of FeCl₂ solutions containing 1, 3, 6, 10, 50, and 100 mM under Ar-statured condition. The value of PZC was determined from the minimum value of the specific capacitance (C_d), which was calculated from the imaginary part (Z”) of the impedance spectra and angular frequency (ω, 1 Hz) according to Eq. 5:

$${\mathrm{C}}_{\mathrm{d}}=\left|1/\left(\omega Z\text{''}\right)\right|.$$ [5]

The OCP of NA-FeOOH was also carried out by the electrochemical workstation (CHI 660e) in 5 mM NaCl solution under Ar-saturated condition.

Redox Potential of Adsorbed Fe(II) Determination.

The ratio of AQDS(red)/AQDS(ox) was determined when the adsorbed Fe(II) reached equilibration. In this case, the reduction potential of E (AQDS) system can be calculated by the Nernst equation (Eqs. 6–8). Once the redox reaction was equilibrated (ΔE = 0), the NA-FeOOH electrode (1 cm²) under different voltages (0, −0.9, −1.1, −1.3, and −1.5 V vs. Ag/AgCl) was taken out separately and rinsed several times with Millipore water. Afterward, the rinsed electrode was quickly placed in a centrifuge tube (10 mL) filled with 5 mL of 500 µM AQDS(ox). With the reaction time reaching under Ar-saturated conditions, the solution became progressively pinker (SI Appendix, Fig. S26). This indicated the generation of AQDS (red) by the adsorbed Fe(II). After stabilization of the solution color (meaning redox reactions of AQDS in the suspensions on the time scale of seconds to minutes), the reaction was stopped and the NA-FeOOH electrode was taken out. Afterward, the absorbance spectra of the remaining solution were recorded using airtight 1 cm quartz cuvettes from 290 to 600 nm using UV-vis spectrophotometry (U3900, HITACHI). The ratio of AQDS (red) and AQDS (ox) was calculated according to the absorbance at the wavelength of ~330 and ~410 nm (SI Appendix, Fig. S27). The electrodes with different reaction times (5, 15, 30, 45, and 60 min) were measured by the same procedure. Controls containing only the NA-FeOOH without applied voltage and Millipore water were also carried out for comparison. All batches were prepared in duplicates to minimize errors, and all experiments and air-sensitive procedures were conducted inside a glovebox (Mikrouna) under Ar-saturated conditions.

$$\mathrm{\Delta }E=E\left(\mathrm{AQDS}\right)-E\left(\mathrm{Fe}\right) (\mathrm{equilibrium}, \mathrm{\Delta }E=0)\mathit{,}$$ [6]

$$E\left(\mathrm{AQDS}\right){=\mathrm{E}}_0\left(\mathrm{AQDS}\right)-\frac{\mathrm{RT}}{2\mathrm{F}}\ln \left\{\frac{\left[\mathrm{AQDS}\left(\mathrm{red}\right)\right]}{\left[\mathrm{AQDS}\left(\mathrm{ox}\right)\right]}\right\},$$ [7]

$$E\left(\mathrm{Fe}\right){=\mathrm{E}}_0\left(\mathrm{Fe}\right)-\frac{\mathrm{RT}}{\mathrm{F}}\ln \left\{\frac{\left[\mathrm{Fe}\left(\mathrm{II}\right)\right]}{\left[\mathrm{Fe}\left(\mathrm{III}\right)\right]}\right\}.$$ [8]

Density Functional Theory Calculations.

The (001) plane of β-FeOOH was built with the vacuum space along the Z-direction set to 18 Å, enough to avoid interaction between the two neighboring images. Next, NO₃⁻ ion was loaded on the surface, the bottom three atomic layers were fixed, and the top three atomic layers were relaxed adequately. The first-principle calculations in the framework of density functional theory were carried out using the Cambridge Sequential Total Energy Package known as CASTEP. The exchange-correlation function under the generalized gradient approximation (GGA) with norm-conserving pseudopotentials and Perdew–Burke–Ernzerhof function was adopted to describe the electron–electron interaction. An energy cutoff of 750 eV was used, and a k-point sampling set of 5×5 × 1 was let to converge. In this case, a force tolerance of 0.01 eV Å⁻¹, an energy tolerance of 5.0×10⁻⁷ eV per atom, and a maximum displacement of 5.0×10⁻⁴ Å were considered. The Gibbs free energy diagrams were estimated by the following equation.

$$∆\mathrm{G}=∆\mathrm{E}+∆\mathrm{ZPE}-\mathrm{T}∆\mathrm{S},$$ [9]

$$\begin{array}{c}∆{\mathrm{G}}_{\mathrm{NO}3}{=\mathrm{E}}_{\mathrm{NO}3*}{+\frac{1}{2}\mathrm{E}}_{\mathrm{H}2}{-\mathrm{E}*-\mathrm{E}}_{\mathrm{gas}\left(\mathrm{HNO}3\right)}+0.75 \mathrm{eV},\hfill \end{array}$$ [10]

where ΔE represents the energy change between the reactant and product obtained from DFT calculations, ΔZPE is the change in zero-point energy, and T and ΔS denote temperature and change in entropy, respectively. Here, T = 300 K is considered. E_H2 and E_{gas (HNO3)} refer to the energies of H₂ and HCO₃ gas, respectively.

MD Simulations.

MD simulations were performed to investigate the distribution of NaNO₃ aqueous solution on the (001) plane of β-FeOOH surface in the presence and absence of Fe(II) ions at the solid–liquid interface. The dimensional size of the (001) plane of β-FeOOH substrate was set to 49.03 Å × 49.03 Å × 10.28 Å. The liquid phase in a dimension of 49.03 Å × 49.03 Å × 60.00 Å contained 226 Na⁺, 226 NO₃⁻, and 4,520 water molecules to form NaNO₃ aqueous solution with a density of 1.23 g cm⁻³ at 25 °C, placed on the β-FeOOH (001) surface. The adsorption density of Fe(II) ion on FeOOH surfaces was estimated to be 1.04 per nm².

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.